Obesity has become epidemic among U.S. children and results in serious negative health consequences over the lifespan. Thus, identifying effective interventions for prevention of obesity in children is paramount. The purpose of the proposed study is to design, implement, and evaluate a culturally appropriate, multi-level trans-community prevention intervention for decreasing obesity risk (measured by BMI percentile) in 3 to 5- year-old rural Hispanic and American Indian children. The primary intervention site will be Head Start centers, where enrolled low-income preschool children spend the majority of their days. The intervention, Child Health Initiative for Lifelong Eating and Exercise (CHILE), will include policy changes to address environmental determinants of healthful eating and physical activity, as well as individual and interpersonal level strategies to promote increased moderate to vigorous physical activity; increased consumption of fruits, vegetables, and whole grains; decreased consumption of sweetened beverages and high-fat foods; and decreased television viewing/screen time. The intervention includes six components: a classroom curriculum for Head Start classrooms; an in-service training program for Head Start teachers and aides; training for Head Start food service staff; a home- and Head Start-based program for families; use of local health care providers as community change agents, advisors, and role models; and a community grocery store program to increase availability and promotion of healthy food options. The study will employ a nested cohort design in which 16 Head Start centers will be randomly assigned to intervention or control conditions. Two cohorts of 3-year-old children within each Head Start center will be followed for two years post-randomization. Participants will be assessed using a measurement and interview protocol across four time points. Data will be analyzed using multi-level modeling to partition variance into individual- and Head Start- or intervention- level components, and to explore individuals' change over time and potential mediating effects. [unreadable] [unreadable]